TIDE engine de-identifies clinical notes preserving patient data links
TIDE 2.0: an open, model-agnostic engine for keyed de-identification of clinical notes
Computation and LanguageArtificial Intelligence
Summary
Clinical notes contain detailed patient information but need personal details removed before research use. The authors present TIDE 2.0, a tool that finds and replaces personal info without losing important connections between patient records. It uses a two-step process: detecting private info and replacing it consistently with coded substitutes that keep time relationships intact. This makes patient data safer to use while preserving research value. The tool is open source and can run on local computers for privacy.
What this means in practice
- •For hospital data teams: Remove patient identifiers from clinical notes while preserving links for internal research and audits using a locally run, configurable engine.
- •For healthcare software developers: Integrate cryptographically keyed anonymization into health record tools to consistently replace patient data without storing linkage tables or losing timeline information.
Authors
Jose D. Posada, Somalee Datta, Priya Desai
Abstract
Clinical notes capture most of what is documented about a patient's care, but they cannot be used for research until protected health information (PHI) is removed. De-identification is often treated as a detection problem. Detection alone is not sufficient: redaction strips clinical content along with identifiers, date blanking destroys the temporal intervals needed for longitudinal analysis, and assigning a fresh random surrogate at each occurrence breaks links between a patient's notes. We present TIDE 2.0, an MIT-licensed engine with two separable stages: an interchangeable recognizer and a keyed anonymizer. Both run on hardware the institution owns. Surrogates are generated cryptographically with no stored linkage table. Dates shift by a per-patient, interval-preserving offset; each value receives the same surrogate across all occurrences under a given key; and a release produced under a new key cannot be linked to earlier releases. We also release TIDE2-Sentry, a recognizer distilled from a large language model. On two gold-annotated corpora from two institutions, the default configuration reached span-level recall of 0.88 in-domain and 0.77 on the second institution's corpus, at precision 0.88 and 0.87. We report recall and precision per category alongside these aggregates. The engine is open source, and the recognizer is available under a gated research-use agreement, so institutions can run, inspect and extend both within their own environments.